HRS 2016 - Cognition classification

Author

Doug Tommet

Published

July 18, 2025

Data

The data used in this analysis comes from the 2016 wave of the HRS study. The data was directly downloaded from their website.

The files used are:

  • trk2022tr_r.dta

  • H16A_R.dta

  • H16D_R.dta

To norm the factor scores, RNJ provided a dataset that identified the participants with normal cognition in the HCAP sample.

Recode demographic variables

This section shows the recoding of the demographic variables.

Age came from the variable PA019 and no recoding was done.

Years of education came from the variable SCHLYRS and no recoding was done. Five participants had their values changed based on RNJ’s code.

Female came from the variable GENDER.

Table 1: Coding of female indicator
Female from trk2022tr_r GENDER n
0 1. Male 8664
1 2. Female 12245
NA NA 3

Black came from the variables RACE and HISPANIC.

Table 2: Coding of Black or African-American indicator
Black or African-American (not Hispanic) from trk2022tr_r RACE HISPANIC n
0 0. Not obtained 0. Not obtained 5
0 0. Not obtained 1. Hispanic, Mexican 40
0 0. Not obtained 2. Hispanic, Other 23
0 0. Not obtained 3. Hispanic, type unknown 1
0 0. Not obtained 5. Non-Hispanic 9
0 1. White/Caucasian 0. Not obtained 9
0 1. White/Caucasian 1. Hispanic, Mexican 1138
0 1. White/Caucasian 2. Hispanic, Other 677
0 1. White/Caucasian 3. Hispanic, type unknown 18
0 1. White/Caucasian 5. Non-Hispanic 11980
0 2. Black or African American 1. Hispanic, Mexican 10
0 2. Black or African American 2. Hispanic, Other 86
0 2. Black or African American 3. Hispanic, type unknown 2
0 7. Other 0. Not obtained 15
0 7. Other 1. Hispanic, Mexican 795
0 7. Other 2. Hispanic, Other 632
0 7. Other 3. Hispanic, type unknown 5
0 7. Other 5. Non-Hispanic 1030
1 2. Black or African American 0. Not obtained 12
1 2. Black or African American 5. Non-Hispanic 4422
NA NA NA 3

Hispanic came from the variable HISPANIC.

Table 3: Coding of Hispanic indicator
Hispanic from trk2022tr_r HISPANIC n
0 0. Not obtained 41
0 5. Non-Hispanic 17441
1 1. Hispanic, Mexican 1983
1 2. Hispanic, Other 1418
1 3. Hispanic, type unknown 26
NA NA 3

Recode the functional impairment items

This section shows the recoding of the IADL items.

There are 11 IADL items that are considered to measure functional impairment.

  • Item responses of 1 (Yes) and 6 (Can’t do) are recoded as “impairment”

  • Item responses of 5 (No), 7 (Don’t do), 8 (Don’t know) and 9 (Refused) are recoded as “no impairment”

Two of the items - “Difficulty taking medication” and “Do you think you would have difficulty taking medication” are logically dependent and are merged into one item

The sum of these 10 recoded IADL items is the functional impairment score

Table 4: ADL: Difficulty Dressing recoding
ADL: Difficulty Dressing
Total
Impaired Not impaired Unknown
DIFFICULTY- DRESSING



    1 2,352 0 0 2,352
    5 0 10,875 0 10,875
    6 52 0 0 52
    7 0 0 12 12
    8 0 0 10 10
    9 0 0 2 2
    Unknown 0 0 7,609 7,609
Total 2,404 10,875 7,633 20,912
Table 5: ADL: Difficulty Bathing recoding
ADL: Difficulty Bathing
Total
Impaired Not impaired Unknown
DIFFICULTY BATHING



    1 1,690 0 0 1,690
    5 0 8,552 0 8,552
    6 39 0 0 39
    7 0 0 6 6
    8 0 0 6 6
    9 0 0 1 1
    Unknown 0 0 10,618 10,618
Total 1,729 8,552 10,631 20,912
Table 6: ADL: Difficulty Eating recoding
ADL: Difficulty Eating
Total
Impaired Not impaired Unknown
DIFFICULTY EATING



    1 806 0 0 806
    5 0 9,439 0 9,439
    6 28 0 0 28
    7 0 0 15 15
    8 0 0 4 4
    9 0 0 2 2
    Unknown 0 0 10,618 10,618
Total 834 9,439 10,639 20,912
Table 7: ADL: Difficulty Using Toilet recoding
ADL: Difficulty Using Toilet
Total
Impaired Not impaired Unknown
DIFFICULTY USING TOILET



    1 1,336 0 0 1,336
    5 0 8,892 0 8,892
    6 32 0 0 32
    7 0 0 23 23
    8 0 0 10 10
    9 0 0 1 1
    Unknown 0 0 10,618 10,618
Total 1,368 8,892 10,652 20,912
Table 8: ADL: Difficulty Using Maps recoding
IADL: Difficulty Using Maps
Total
Impaired Not impaired Unknown
DIFFICULTY- USING MAPS



    1 2,517 0 0 2,517
    5 0 16,811 0 16,811
    6 268 0 0 268
    7 0 0 1,233 1,233
    8 0 0 39 39
    9 0 0 7 7
    Unknown 0 0 37 37
Total 2,785 16,811 1,316 20,912
Table 9: IADL: Difficulty Meal Prep recoding
IADL: Difficulty Meal Prep
Total
Impaired Not impaired Unknown
IADL MEAL PREPARATION DIFFICULTY



    1 1,207 0 0 1,207
    5 0 18,880 0 18,880
    6 101 0 0 101
    7 0 0 679 679
    8 0 0 5 5
    9 0 0 3 3
    Unknown 0 0 37 37
Total 1,308 18,880 724 20,912
Table 10: IADL: Difficulty Grocery Shopping recoding
IADL: Difficulty Grocery Shopping
Total
Impaired Not impaired Unknown
IADL GROC SHOP DIFFICULTY



    1 1,761 0 0 1,761
    5 0 18,324 0 18,324
    6 81 0 0 81
    7 0 0 701 701
    8 0 0 6 6
    9 0 0 2 2
    Unknown 0 0 37 37
Total 1,842 18,324 746 20,912
Table 11: IADL: Difficulty Making Phone Calls recoding
IADL: Difficulty Making Phone Calls
Total
Impaired Not impaired Unknown
IADL MAKING PHONE CALLS DIFFICULTY



    1 888 0 0 888
    5 0 19,782 0 19,782
    6 51 0 0 51
    7 0 0 146 146
    8 0 0 4 4
    9 0 0 4 4
    Unknown 0 0 37 37
Total 939 19,782 191 20,912
Table 12: IADL: Difficulty Taking Meds recoding
IADL: Difficulty Taking Meds IADL TAKING MEDICATION DIFFICULTY IADL TAKING MEDS IF NEEDED DIFFICULTY n
0 5 NA 19671
0 7 5 202
1 1 NA 923
1 6 NA 24
1 7 1 41
NA 7 8 6
NA 8 NA 5
NA 9 NA 3
NA NA NA 37
Table 13: IADL: Difficulty Managing Money Calls recoding
IADL: Difficulty Managing Money
Total
Impaired Not impaired Unknown
IADL MANAGING MONEY DIFFICULTY



    1 1,339 0 0 1,339
    5 0 18,751 0 18,751
    6 53 0 0 53
    7 0 0 708 708
    8 0 0 17 17
    9 0 0 7 7
    Unknown 0 0 37 37
Total 1,392 18,751 769 20,912
Table 14: Descriptive statistics of recoded functional impairment items
Characteristic N = 20,9121
ADL: Difficulty Dressing
    Impaired 2,404 (18%)
    Not impaired 10,875 (82%)
    Unknown 7,633
ADL: Difficulty Bathing
    Impaired 1,729 (17%)
    Not impaired 8,552 (83%)
    Unknown 10,631
ADL: Difficulty Eating
    Impaired 834 (8.1%)
    Not impaired 9,439 (92%)
    Unknown 10,639
ADL: Difficulty Using Toilet
    Impaired 1,368 (13%)
    Not impaired 8,892 (87%)
    Unknown 10,652
IADL: Difficulty Using Maps
    Impaired 2,785 (14%)
    Not impaired 16,811 (86%)
    Unknown 1,316
IADL: Difficulty Meal Prep
    Impaired 1,308 (6.5%)
    Not impaired 18,880 (94%)
    Unknown 724
IADL: Difficulty Grocery Shopping
    Impaired 1,842 (9.1%)
    Not impaired 18,324 (91%)
    Unknown 746
IADL: Difficulty Making Phone Calls
    Impaired 939 (4.5%)
    Not impaired 19,782 (95%)
    Unknown 191
IADL: Difficulty Taking Meds
    Impaired 988 (4.7%)
    Not impaired 19,873 (95%)
    Unknown 51
IADL: Difficulty Managing Money
    Impaired 1,392 (6.9%)
    Not impaired 18,751 (93%)
    Unknown 769
Sum of ADL/IADL impairments
    0 14,821 (71%)
    1 2,946 (14%)
    2 1,033 (4.9%)
    3 651 (3.1%)
    4 446 (2.1%)
    5 304 (1.5%)
    6 244 (1.2%)
    7 151 (0.7%)
    8 131 (0.6%)
    9 99 (0.5%)
    10 86 (0.4%)
1 n (%)

Recode the informant based items

This section shows the recoding of the Jorm items and the subjective cognitive impairment item.

There are 16 Jorm items.

  • For each Jorm item, three HRS items need to be combined to recreate the Jorm item

The average of these 16 Jorms items is the score

The subjective cognitive impairment item was recoded so that “no change” and “improvement” was combined into one category.

Table 15: Jorm: Remembering things about family
Remembering things about family RATE R AT REMEMBERING THINGS- PC ORGANIZATION IMPROVED- PC ORGANIZATION WORSE- PC n
1 1 1 NA 8
2 1 2 NA 8
3 2 NA NA 377
4 3 NA 4 111
5 3 NA 5 213
NA 3 NA 8 2
NA 4 NA NA 2
NA NA NA NA 20191
Table 16: Jorm: Remembering things that happened recently
Remembering things that happened recently RATE R AT REMEMBERING RECENT EVENTS- PC REMEMBERING RECENT EVENTS IMPROVED- PC REMEMBERING RECENT EVENTS WORSE- PC n
1 1 1 NA 8
2 1 2 NA 6
3 2 NA NA 352
4 3 NA 4 116
5 3 NA 5 234
NA 3 NA 8 1
NA 4 NA NA 3
NA 9 NA NA 1
NA NA NA NA 20191
Table 17: Jorm: Recalling conversations a few day later
Recalling conversations a few day later RATE R AT CONVERSATION RECALL- PC CONVERSATION RECALL IMPROVED- PC CONVERSATION RECALL WORSE- PC n
1 1 1 NA 6
2 1 2 NA 4
3 2 NA NA 353
4 3 NA 4 124
5 3 NA 5 222
NA 1 9 NA 1
NA 4 NA NA 10
NA 8 NA NA 1
NA NA NA NA 20191
Table 18: Jorm: Remembering telephone number
Remembering telephone number RATE REMEMBERING OWN PHONE NUM- PC REMEMBERING OWN PHONE NUM IMPROVE- PC REMEMBERING OWN PHONE NUM WORSE- PC n
1 1 1 NA 7
2 1 2 NA 4
3 2 NA NA 425
4 3 NA 4 58
5 3 NA 5 192
NA 4 NA NA 31
NA 8 NA NA 3
NA 9 NA NA 1
NA NA NA NA 20191
Table 19: Jorm: Remembering day and month
Remembering day and month RATE REMEMBERING CURRENT DY/MO- PC REMEMBERING CURRENT DY/MO IMPROVE- PC REMEMBERING CURRENT DY/MO WORSE- PC n
1 1 1 NA 5
2 1 2 NA 4
3 2 NA NA 391
4 3 NA 4 91
5 3 NA 5 206
NA 4 NA NA 21
NA 8 NA NA 2
NA 9 NA NA 1
NA NA NA NA 20191
Table 20: Jorm: Remembering where things are kept
Remembering where things are kept RATE REMEMBERING WHERE THINGS KEPT- PC WHERE THINGS ARE KEPT IMPROVED- PC WHERE THINGS ARE KEPT WORSE- PC n
1 1 1 NA 3
2 1 2 NA 4
3 2 NA NA 415
4 3 NA 4 91
5 3 NA 5 185
NA 4 NA NA 20
NA 8 NA NA 1
NA 9 NA NA 2
NA NA NA NA 20191
Table 21: Jorm: Remembering where to find things
Remembering where to find things RATE FINDING THINGS IN DIFF PLACES- PC FINDING THINGS IMPROVED- PC FINDING THINGS WORSE- PC n
1 1 1 NA 5
2 1 2 NA 6
3 2 NA NA 343
4 3 NA 4 113
5 3 NA 5 218
NA 1 9 NA 1
NA 3 NA 8 1
NA 4 NA NA 29
NA 8 NA NA 4
NA 9 NA NA 1
NA NA NA NA 20191
Table 22: Jorm: Knowing how to work familar machines around the house
Knowing how to work familar machines around the house RATE WORKING WITH FAMILIAR MACHINES- PC WORKING WITH FAMILIAR MACHINES IMPR- PC WORKING WITH FAMILIAR MACHINES WORSE- PC n
1 1 1 NA 6
2 1 2 NA 6
3 2 NA NA 371
4 3 NA 4 57
5 3 NA 5 167
NA 4 NA NA 111
NA 9 NA NA 3
NA NA NA NA 20191
Table 23: Jorm: Learning to use a new gadget
Learning to use a new gadget RATE LEARNING NEW MACHINES- PC LEARNING NEW MACHINES IMPROVED- PC LEARNING NEW MACHINES WORSE- PC n
1 1 1 NA 6
2 1 2 NA 8
3 2 NA NA 317
4 3 NA 4 60
5 3 NA 5 198
NA 4 NA NA 130
NA 9 NA NA 2
NA NA NA NA 20191
Table 24: Jorm: Learning new things in general
Learning new things in general RATE LEARNING NEW THINGS IN GENERAL- PC LEARNING ABILITY IMPROVE- PC LEARNING ABILITY WORSE- PC n
1 1 1 NA 12
2 1 2 NA 9
3 2 NA 4 1
3 2 NA NA 349
4 3 NA 4 73
5 3 NA 5 214
NA 3 NA 8 1
NA 4 NA NA 57
NA 8 NA NA 2
NA 9 NA NA 3
NA NA NA NA 20191
Table 25: Jorm: Following a story in a book or on TV
Following a story in a book or on TV RATE ABILITY TO FOLLOW STORY- PC ABILITY TO FOLLOW STORY IMPROVE- PC ABILITY TO FOLLOW STORY WORSE- PC n
1 1 1 NA 13
2 1 2 NA 6
3 2 NA NA 403
4 3 NA 4 77
5 3 NA 5 175
NA 1 9 NA 1
NA 3 NA 8 1
NA 4 NA NA 37
NA 8 NA NA 6
NA 9 NA NA 2
NA NA NA NA 20191
Table 26: Jorm: Making decisions on everyday matters
Making decisions on everyday matters RATE MAKING DECISIONS- PC MAKE DECISIONS IMPROVE- PC MAKE DECISIONS WORSE- PC n
1 1 1 NA 7
2 1 2 NA 6
3 2 NA NA 380
4 3 NA 4 85
5 3 NA 5 186
NA 4 NA NA 55
NA 9 NA NA 2
NA NA NA NA 20191
Table 27: Jorm: Handling money for shopping
Handling money for shopping RATE HANDLING SHOPPING MONEY- PC HANDLING SHOPPING MONEY IMPROVE- PC HANDLING SHOPPING MONEY WORSE- PC n
1 1 1 NA 6
2 1 2 NA 3
3 2 NA NA 335
4 3 NA 4 37
5 3 NA 5 141
NA 4 NA NA 198
NA 9 NA NA 1
NA NA NA NA 20191
Table 28: Jorm: Handling financial matters
Handling financial matters RATE HANDLING FINANCES- PC HANDLING FINANCES IMPROVE- PC HANDLING FINANCES WORSE- PC n
1 1 1 NA 5
2 1 2 NA 6
3 2 NA NA 289
4 3 NA 4 32
5 3 NA 5 139
NA 4 NA NA 249
NA 9 NA NA 1
NA NA NA NA 20191
Table 29: Jorm: Handling everyday arthimetic problems
Handling everyday arthimetic problems RATE HANDLING DAILY ARITHMETIC PROBS- PC HANDLING ARITHMETIC PROBLEMS IMPROVE- PC HANDLING ARITHMETIC PROBLEMS WORSE- PC n
1 1 1 NA 8
2 1 2 NA 6
3 2 NA NA 357
4 3 NA 4 47
5 3 NA 5 178
NA 4 NA NA 122
NA 8 NA NA 2
NA 9 NA NA 1
NA NA NA NA 20191
Table 30: Jorm: Using intelligence to understand what’s going on
Using intelligence to understand what's going on RATE REASONING- PC REASONING IMPROVE- PC REASONING WORSE- PC n
1 1 1 NA 7
2 1 2 NA 12
3 2 NA NA 384
4 3 NA 4 92
5 3 NA 5 189
NA 3 NA 8 1
NA 4 NA NA 32
NA 8 NA NA 3
NA 9 NA NA 1
NA NA NA NA 20191
Table 31: Jorm Score
Characteristic N = 20,912
Jorm score (HRS)
    Mean (SD) 3.72 (0.83)
    Median (Q1, Q3) 3.38 (3.00, 4.56)
    N Non-missing 721
    Unknown 20,191
Jorm score (HRS HCAP variable set 3)
    Mean (SD) 3.18 (0.55)
    Median (Q1, Q3) 3.04 (3.00, 3.31)
    N Non-missing 3,496
    Unknown 17,416
Figure 1: Distribution of Jorm scores in HRS & HCAP
Figure 2: Scatterplot of Jorm scores by HRS & HCAP
Table 32: Self-concerns
Characteristic N = 20,9121
Compared to two years ago, would you say your memory is better now, about the same, or worse now than it was then?
    Worse 4,315 / 19,938 (22%)
    Same/Better 15,623 / 19,938 (78%)
    Unknown 974
1 n / N (%)

Recode cognitive items

This section shows the recoding of the cognitive items.

In general, I tried to have higher values indicate correct or better performance.

I recoded “don’t know” and “refusal” to incorrect. These could be changed.

If all the component values of a vd variable are NA, then I set the vd variable to NA.

The tables show the recoded variable on the left column, the original variable in the center, and the sample size on the right column.

Orientation

The number correct of the four orientation to time questions - Month, Day, Year and Day of week.

Table 33: vdori - Orientation to time - number correct recoding
Orientation to Time - number correct PD151 PD152 PD153 PD154 n
0 5. MONTH NOT OK 5. DAY OF MONTH NOT OK 5. YEAR NOT OK 5. DAY NOT OK 34
0 5. MONTH NOT OK 5. DAY OF MONTH NOT OK 5. YEAR NOT OK 8. DK (Don't Know) 2
0 5. MONTH NOT OK 5. DAY OF MONTH NOT OK 8. DK (Don't Know) 5. DAY NOT OK 6
0 5. MONTH NOT OK 5. DAY OF MONTH NOT OK 8. DK (Don't Know) 8. DK (Don't Know) 1
0 5. MONTH NOT OK 8. DK (Don't Know) 5. YEAR NOT OK 5. DAY NOT OK 1
0 5. MONTH NOT OK 8. DK (Don't Know) 5. YEAR NOT OK 8. DK (Don't Know) 1
0 5. MONTH NOT OK 8. DK (Don't Know) 8. DK (Don't Know) 5. DAY NOT OK 5
0 5. MONTH NOT OK 8. DK (Don't Know) 8. DK (Don't Know) 8. DK (Don't Know) 2
0 8. DK (Don't Know) 5. DAY OF MONTH NOT OK 5. YEAR NOT OK 5. DAY NOT OK 2
0 8. DK (Don't Know) 5. DAY OF MONTH NOT OK 5. YEAR NOT OK 8. DK (Don't Know) 1
0 8. DK (Don't Know) 5. DAY OF MONTH NOT OK 8. DK (Don't Know) 5. DAY NOT OK 1
0 8. DK (Don't Know) 8. DK (Don't Know) 5. YEAR NOT OK 5. DAY NOT OK 4
0 8. DK (Don't Know) 8. DK (Don't Know) 5. YEAR NOT OK 8. DK (Don't Know) 1
0 8. DK (Don't Know) 8. DK (Don't Know) 8. DK (Don't Know) 5. DAY NOT OK 16
0 8. DK (Don't Know) 8. DK (Don't Know) 8. DK (Don't Know) 8. DK (Don't Know) 46
0 9. RF (Refused) 9. RF (Refused) 9. RF (Refused) 5. DAY NOT OK 1
0 9. RF (Refused) 9. RF (Refused) 9. RF (Refused) 9. RF (Refused) 14
1 1. MONTH OK 5. DAY OF MONTH NOT OK 5. YEAR NOT OK 5. DAY NOT OK 35
1 1. MONTH OK 5. DAY OF MONTH NOT OK 5. YEAR NOT OK 8. DK (Don't Know) 1
1 1. MONTH OK 5. DAY OF MONTH NOT OK 8. DK (Don't Know) 5. DAY NOT OK 3
1 1. MONTH OK 5. DAY OF MONTH NOT OK 8. DK (Don't Know) 8. DK (Don't Know) 1
1 1. MONTH OK 5. DAY OF MONTH NOT OK 9. RF (Refused) 5. DAY NOT OK 1
1 1. MONTH OK 8. DK (Don't Know) 5. YEAR NOT OK 5. DAY NOT OK 4
1 1. MONTH OK 8. DK (Don't Know) 5. YEAR NOT OK 8. DK (Don't Know) 3
1 1. MONTH OK 8. DK (Don't Know) 8. DK (Don't Know) 5. DAY NOT OK 3
1 1. MONTH OK 8. DK (Don't Know) 8. DK (Don't Know) 8. DK (Don't Know) 7
1 5. MONTH NOT OK 1. DAY OF MONTH OK 5. YEAR NOT OK 5. DAY NOT OK 5
1 5. MONTH NOT OK 1. DAY OF MONTH OK 8. DK (Don't Know) 8. DK (Don't Know) 2
1 5. MONTH NOT OK 5. DAY OF MONTH NOT OK 1. YEAR OK 5. DAY NOT OK 12
1 5. MONTH NOT OK 5. DAY OF MONTH NOT OK 1. YEAR OK 8. DK (Don't Know) 2
1 5. MONTH NOT OK 5. DAY OF MONTH NOT OK 5. YEAR NOT OK 1. DAY OK 39
1 5. MONTH NOT OK 5. DAY OF MONTH NOT OK 8. DK (Don't Know) 1. DAY OK 4
1 5. MONTH NOT OK 8. DK (Don't Know) 1. YEAR OK 5. DAY NOT OK 1
1 5. MONTH NOT OK 8. DK (Don't Know) 1. YEAR OK 8. DK (Don't Know) 1
1 5. MONTH NOT OK 8. DK (Don't Know) 5. YEAR NOT OK 1. DAY OK 3
1 5. MONTH NOT OK 8. DK (Don't Know) 8. DK (Don't Know) 1. DAY OK 6
1 8. DK (Don't Know) 1. DAY OF MONTH OK 5. YEAR NOT OK 5. DAY NOT OK 1
1 8. DK (Don't Know) 1. DAY OF MONTH OK 8. DK (Don't Know) 8. DK (Don't Know) 1
1 8. DK (Don't Know) 5. DAY OF MONTH NOT OK 1. YEAR OK 8. DK (Don't Know) 1
1 8. DK (Don't Know) 5. DAY OF MONTH NOT OK 8. DK (Don't Know) 1. DAY OK 1
1 8. DK (Don't Know) 8. DK (Don't Know) 1. YEAR OK 5. DAY NOT OK 2
1 8. DK (Don't Know) 8. DK (Don't Know) 1. YEAR OK 8. DK (Don't Know) 2
1 8. DK (Don't Know) 8. DK (Don't Know) 5. YEAR NOT OK 1. DAY OK 4
1 8. DK (Don't Know) 8. DK (Don't Know) 8. DK (Don't Know) 1. DAY OK 26
2 1. MONTH OK 1. DAY OF MONTH OK 5. YEAR NOT OK 5. DAY NOT OK 26
2 1. MONTH OK 1. DAY OF MONTH OK 5. YEAR NOT OK 8. DK (Don't Know) 2
2 1. MONTH OK 1. DAY OF MONTH OK 8. DK (Don't Know) 5. DAY NOT OK 2
2 1. MONTH OK 1. DAY OF MONTH OK 8. DK (Don't Know) 8. DK (Don't Know) 1
2 1. MONTH OK 5. DAY OF MONTH NOT OK 1. YEAR OK 5. DAY NOT OK 99
2 1. MONTH OK 5. DAY OF MONTH NOT OK 1. YEAR OK 8. DK (Don't Know) 10
2 1. MONTH OK 5. DAY OF MONTH NOT OK 5. YEAR NOT OK 1. DAY OK 71
2 1. MONTH OK 5. DAY OF MONTH NOT OK 8. DK (Don't Know) 1. DAY OK 12
2 1. MONTH OK 8. DK (Don't Know) 1. YEAR OK 5. DAY NOT OK 9
2 1. MONTH OK 8. DK (Don't Know) 1. YEAR OK 8. DK (Don't Know) 6
2 1. MONTH OK 8. DK (Don't Know) 5. YEAR NOT OK 1. DAY OK 6
2 1. MONTH OK 8. DK (Don't Know) 8. DK (Don't Know) 1. DAY OK 7
2 5. MONTH NOT OK 1. DAY OF MONTH OK 1. YEAR OK 5. DAY NOT OK 4
2 5. MONTH NOT OK 1. DAY OF MONTH OK 5. YEAR NOT OK 1. DAY OK 10
2 5. MONTH NOT OK 1. DAY OF MONTH OK 8. DK (Don't Know) 1. DAY OK 2
2 5. MONTH NOT OK 5. DAY OF MONTH NOT OK 1. YEAR OK 1. DAY OK 111
2 5. MONTH NOT OK 8. DK (Don't Know) 1. YEAR OK 1. DAY OK 7
2 8. DK (Don't Know) 1. DAY OF MONTH OK 1. YEAR OK 8. DK (Don't Know) 1
2 8. DK (Don't Know) 1. DAY OF MONTH OK 5. YEAR NOT OK 1. DAY OK 3
2 8. DK (Don't Know) 1. DAY OF MONTH OK 8. DK (Don't Know) 1. DAY OK 3
2 8. DK (Don't Know) 5. DAY OF MONTH NOT OK 1. YEAR OK 1. DAY OK 9
2 8. DK (Don't Know) 8. DK (Don't Know) 1. YEAR OK 1. DAY OK 19
3 1. MONTH OK 1. DAY OF MONTH OK 1. YEAR OK 5. DAY NOT OK 189
3 1. MONTH OK 1. DAY OF MONTH OK 1. YEAR OK 8. DK (Don't Know) 15
3 1. MONTH OK 1. DAY OF MONTH OK 5. YEAR NOT OK 1. DAY OK 136
3 1. MONTH OK 1. DAY OF MONTH OK 8. DK (Don't Know) 1. DAY OK 16
3 1. MONTH OK 5. DAY OF MONTH NOT OK 1. YEAR OK 1. DAY OK 1834
3 1. MONTH OK 8. DK (Don't Know) 1. YEAR OK 1. DAY OK 137
3 1. MONTH OK 9. RF (Refused) 1. YEAR OK 1. DAY OK 1
3 5. MONTH NOT OK 1. DAY OF MONTH OK 1. YEAR OK 1. DAY OK 144
3 8. DK (Don't Know) 1. DAY OF MONTH OK 1. YEAR OK 1. DAY OK 9
4 1. MONTH OK 1. DAY OF MONTH OK 1. YEAR OK 1. DAY OK 10490
NA NA NA NA NA 7212
Table 34: vdori - Orientation to time - number correct
Characteristic N = 20,9121
Orientation to Time - number correct
    0 138 (1.0%)
    1 171 (1.2%)
    2 420 (3.1%)
    3 2,481 (18%)
    4 10,490 (77%)
    Unknown 7,212
1 n (%)

Count backwards

Table 35: vdcount: Count backwards recoding
Count backwards from 20 PD124 PD129 n
0 5. INCORRECT NA 1324
0 6. WANTS TO START OVER 5. INCORRECT 11
0 6. WANTS TO START OVER 9. RF (Refused) 2
0 9. RF (Refused) NA 20
1 1. CORRECT NA 18519
1 6. WANTS TO START OVER 1. CORRECT 35
NA NA NA 1001
Table 36: vdcount: Count backwards
Characteristic N = 20,9121
Count backwards from 20
    Correct 18,554 (93%)
    Incorrect 1,357 (6.8%)
    Unknown 1,001
1 n (%)

Object naming

The scissors and cactus naming items have been combined into one item.

Table 37: vdlfl2: Scissors & Cactus - number correct recoding
Object naming - Scissors, cactus PD155 PD156 n
0 5. NOT CORRECT 5. NOT CORRECT 20
0 5. NOT CORRECT 8. DK (Don't Know) 20
0 5. NOT CORRECT 9. RF (Refused) 1
0 8. DK (Don't Know) 5. NOT CORRECT 9
0 8. DK (Don't Know) 8. DK (Don't Know) 27
0 9. RF (Refused) 9. RF (Refused) 17
1 1. SCISSORS OR SHEARS ONLY 5. NOT CORRECT 367
1 1. SCISSORS OR SHEARS ONLY 8. DK (Don't Know) 643
1 1. SCISSORS OR SHEARS ONLY 9. RF (Refused) 4
1 5. NOT CORRECT 1. CACTUS OR NAME OF KIND OF CACTUS 99
1 8. DK (Don't Know) 1. CACTUS OR NAME OF KIND OF CACTUS 28
1 9. RF (Refused) 1. CACTUS OR NAME OF KIND OF CACTUS 3
2 1. SCISSORS OR SHEARS ONLY 1. CACTUS OR NAME OF KIND OF CACTUS 12462
NA NA NA 7212
Table 38: vdlfl2: Scissors & Cactus - number correct
Characteristic N = 20,9121
Object naming - Scissors, cactus
    0 94 (0.7%)
    1 1,144 (8.4%)
    2 12,462 (91%)
    Unknown 7,212
1 n (%)

President/Vice-president

The president and vice-president naming items have been combined into one item.

Table 39: vdlfl3: President and vice-president - number correct recoding
Naming - President, Vice-president PD157 PD158 n
0 5. NOT CORRECT 5. NOT CORRECT 51
0 5. NOT CORRECT 8. DK (Don't Know) 45
0 8. DK (Don't Know) 5. NOT CORRECT 11
0 8. DK (Don't Know) 8. DK (Don't Know) 241
0 8. DK (Don't Know) 9. RF (Refused) 1
0 9. RF (Refused) 8. DK (Don't Know) 4
0 9. RF (Refused) 9. RF (Refused) 23
1 1. LAST NAME CORRECT 5. NOT CORRECT 1343
1 1. LAST NAME CORRECT 8. DK (Don't Know) 3347
1 1. LAST NAME CORRECT 9. RF (Refused) 15
1 5. NOT CORRECT 1. LAST NAME CORRECT 14
1 8. DK (Don't Know) 1. LAST NAME CORRECT 15
1 9. RF (Refused) 1. LAST NAME CORRECT 4
2 1. LAST NAME CORRECT 1. LAST NAME CORRECT 8585
NA NA NA 7213
Table 40: vdlfl3: President and vice-president - number correct
Characteristic N = 20,9121
Naming - President, Vice-president
    0 376 (2.7%)
    1 4,738 (35%)
    2 8,585 (63%)
    Unknown 7,213
1 n (%)

Serial 7

The new variable counts the number of correct responses on the serial 7. Each item was checked so that the difference between consecutive items is 7. So if a mistake was made early on, there is an opportunity to get the remaining items correct.

Table 41: PD142-PD146: Serial 7 recoding
Serial 7's - Number correct rPD142 rPD143 rPD144 rPD145 rPD146 n
0 0. Incorrect 0. Incorrect 0. Incorrect 0. Incorrect 0. Incorrect 406
0 0. Incorrect 0. Incorrect 0. Incorrect 0. Incorrect NA 43
0 0. Incorrect 0. Incorrect 0. Incorrect NA NA 116
0 0. Incorrect 0. Incorrect NA NA NA 266
0 0. Incorrect NA NA NA NA 1014
1 0. Incorrect 0. Incorrect 0. Incorrect 0. Incorrect 1. Correct 58
1 0. Incorrect 0. Incorrect 0. Incorrect 1. Correct 0. Incorrect 69
1 0. Incorrect 0. Incorrect 1. Correct 0. Incorrect 0. Incorrect 68
1 0. Incorrect 0. Incorrect 1. Correct 0. Incorrect NA 13
1 0. Incorrect 1. Correct 0. Incorrect 0. Incorrect 0. Incorrect 91
1 0. Incorrect 1. Correct 0. Incorrect 0. Incorrect NA 14
1 0. Incorrect 1. Correct 0. Incorrect NA NA 31
1 1. Correct 0. Incorrect 0. Incorrect 0. Incorrect 0. Incorrect 698
1 1. Correct 0. Incorrect 0. Incorrect 0. Incorrect NA 82
1 1. Correct 0. Incorrect 0. Incorrect NA NA 235
1 1. Correct 0. Incorrect NA NA NA 754
2 0. Incorrect 0. Incorrect 0. Incorrect 1. Correct 1. Correct 36
2 0. Incorrect 0. Incorrect 1. Correct 0. Incorrect 1. Correct 48
2 0. Incorrect 0. Incorrect 1. Correct 1. Correct 0. Incorrect 49
2 0. Incorrect 1. Correct 0. Incorrect 0. Incorrect 1. Correct 24
2 0. Incorrect 1. Correct 0. Incorrect 1. Correct 0. Incorrect 42
2 0. Incorrect 1. Correct 1. Correct 0. Incorrect 0. Incorrect 46
2 0. Incorrect 1. Correct 1. Correct 0. Incorrect NA 18
2 1. Correct 0. Incorrect 0. Incorrect 0. Incorrect 1. Correct 295
2 1. Correct 0. Incorrect 0. Incorrect 1. Correct 0. Incorrect 266
2 1. Correct 0. Incorrect 1. Correct 0. Incorrect 0. Incorrect 249
2 1. Correct 0. Incorrect 1. Correct 0. Incorrect NA 52
2 1. Correct 1. Correct 0. Incorrect 0. Incorrect 0. Incorrect 327
2 1. Correct 1. Correct 0. Incorrect 0. Incorrect NA 67
2 1. Correct 1. Correct 0. Incorrect NA NA 179
3 0. Incorrect 0. Incorrect 1. Correct 1. Correct 1. Correct 72
3 0. Incorrect 1. Correct 0. Incorrect 1. Correct 1. Correct 36
3 0. Incorrect 1. Correct 1. Correct 0. Incorrect 1. Correct 26
3 0. Incorrect 1. Correct 1. Correct 1. Correct 0. Incorrect 52
3 1. Correct 0. Incorrect 0. Incorrect 1. Correct 1. Correct 553
3 1. Correct 0. Incorrect 1. Correct 0. Incorrect 1. Correct 274
3 1. Correct 0. Incorrect 1. Correct 1. Correct 0. Incorrect 633
3 1. Correct 1. Correct 0. Incorrect 0. Incorrect 1. Correct 333
3 1. Correct 1. Correct 0. Incorrect 1. Correct 0. Incorrect 389
3 1. Correct 1. Correct 1. Correct 0. Incorrect 0. Incorrect 264
3 1. Correct 1. Correct 1. Correct 0. Incorrect NA 44
4 0. Incorrect 1. Correct 1. Correct 1. Correct 1. Correct 107
4 1. Correct 0. Incorrect 1. Correct 1. Correct 1. Correct 1417
4 1. Correct 1. Correct 0. Incorrect 1. Correct 1. Correct 871
4 1. Correct 1. Correct 1. Correct 0. Incorrect 1. Correct 493
4 1. Correct 1. Correct 1. Correct 1. Correct 0. Incorrect 851
5 1. Correct 1. Correct 1. Correct 1. Correct 1. Correct 7899
NA NA NA NA NA NA 942
Table 42: PD142-PD146: Serial 7
Characteristic N = 20,9121
Serial 7's - Number correct
    0 1,845 (9.2%)
    1 2,113 (11%)
    2 1,698 (8.5%)
    3 2,676 (13%)
    4 3,739 (19%)
    5 7,899 (40%)
    Unknown 942
1 n (%)

Animal naming

This variable is the number of correct responses minus the number of errors. Missing values are set at the median value of errors.

Table 43: PD196: Animal naming recoding
Animal naming (correct - errors) TOTAL ANIMALS ANSWERS ANIMAL MISTAKES NUMBER n
0 0 NA 12
0 1 3 1
0 1 NA 18
0 2 2 1
0 3 3 2
0 6 7 1
0 7 18 1
0 8 15 1
0 8 16 1
0 9 10 2
0 10 10 1
0 10 11 1
0 11 11 1
0 12 15 1
0 13 15 1
0 13 24 1
0 14 14 1
0 14 17 1
0 15 15 1
0 15 22 1
0 16 22 1
0 17 21 1
0 17 24 1
0 17 29 1
0 17 87 1
0 18 20 1
0 18 22 1
0 20 25 1
0 20 37 1
0 21 55 1
0 22 28 1
0 25 38 1
0 25 42 1
0 25 54 1
0 26 42 1
0 26 93 1
0 30 45 1
0 40 45 1
1 2 1 2
1 2 NA 26
1 4 3 1
1 6 5 2
1 9 8 1
1 11 10 1
1 17 16 1
1 20 19 1
2 2 0 1
2 3 1 5
2 3 NA 54
2 4 2 4
2 5 3 2
2 6 4 1
2 7 5 1
2 8 6 1
2 17 15 1
3 4 1 8
3 4 NA 89
3 5 2 6
3 8 5 1
3 9 6 1
3 11 8 1
3 13 10 1
3 23 20 1
3 28 25 1
4 5 1 12
4 5 NA 131
4 6 2 4
4 7 3 5
4 8 4 7
4 9 5 2
4 14 10 1
5 6 1 22
5 6 NA 210
5 7 2 10
5 8 3 3
5 9 4 1
5 10 5 3
5 11 6 4
5 12 7 1
5 13 8 1
5 15 10 1
5 42 37 1
6 6 0 3
6 7 1 36
6 7 NA 258
6 8 2 14
6 9 3 9
6 10 4 3
6 11 5 8
6 12 6 1
6 13 7 1
6 18 12 1
6 24 18 1
7 8 1 51
7 8 NA 303
7 9 2 17
7 10 3 11
7 11 4 3
7 12 5 7
7 13 6 1
7 15 8 1
7 19 12 1
7 24 17 1
8 8 0 2
8 9 1 75
8 9 NA 378
8 10 2 16
8 11 3 17
8 12 4 8
8 13 5 1
8 14 6 2
8 15 7 1
8 16 8 2
8 20 12 1
8 23 15 1
9 9 0 2
9 10 1 56
9 10 NA 368
9 11 2 43
9 12 3 20
9 13 4 12
9 14 5 7
9 17 8 1
9 19 10 1
9 20 11 1
10 10 0 4
10 11 1 86
10 11 NA 447
10 12 2 48
10 13 3 18
10 14 4 8
10 15 5 5
10 16 6 2
10 20 10 2
10 35 25 1
11 11 0 3
11 12 1 128
11 12 98 1
11 12 NA 431
11 13 2 57
11 14 3 16
11 15 4 11
11 16 5 6
11 17 6 3
11 18 7 1
12 12 0 5
12 13 1 130
12 13 NA 436
12 14 2 55
12 15 3 38
12 16 4 20
12 17 5 9
12 18 6 3
12 19 7 1
13 13 0 7
13 14 1 136
13 14 NA 418
13 15 2 72
13 16 3 33
13 17 4 12
13 18 5 9
13 19 6 3
13 20 7 1
13 22 9 1
13 25 12 1
14 14 0 13
14 15 1 150
14 15 NA 455
14 16 2 75
14 17 3 38
14 18 4 16
14 19 5 9
14 20 6 2
14 21 7 1
14 22 8 1
14 24 10 3
15 15 0 8
15 16 1 151
15 16 NA 464
15 17 2 72
15 18 3 29
15 19 4 10
15 20 5 7
15 21 6 3
15 22 7 1
16 16 0 8
16 17 1 149
16 17 98 1
16 17 NA 467
16 18 2 83
16 19 3 29
16 20 4 13
16 21 5 3
16 22 6 3
17 17 0 6
17 18 1 149
17 18 NA 398
17 19 2 76
17 20 3 34
17 21 4 14
17 22 5 6
18 18 0 11
18 19 1 140
18 19 NA 412
18 20 2 79
18 21 3 27
18 22 4 12
18 23 5 4
18 24 6 1
18 26 8 3
18 27 9 1
19 19 0 10
19 20 1 133
19 20 NA 396
19 21 2 79
19 22 3 30
19 23 4 12
19 24 5 4
19 25 6 2
19 31 12 1
20 20 0 8
20 21 1 131
20 21 NA 396
20 22 2 67
20 23 3 19
20 24 4 8
20 25 5 3
20 26 6 2
20 28 8 1
21 21 0 6
21 22 1 100
21 22 NA 347
21 23 2 53
21 24 3 16
21 25 4 8
21 26 5 4
22 22 0 10
22 23 1 100
22 23 NA 295
22 24 2 46
22 25 3 27
22 26 4 7
22 27 5 2
22 36 14 1
23 23 0 7
23 24 1 104
23 24 NA 276
23 25 2 39
23 26 3 14
23 27 4 7
23 28 5 3
23 30 7 1
24 24 0 3
24 25 1 79
24 25 NA 262
24 26 2 43
24 27 3 19
24 28 4 2
24 29 5 2
24 31 7 1
25 25 0 2
25 26 1 64
25 26 NA 187
25 27 2 30
25 28 3 12
25 29 4 8
25 33 8 1
26 26 0 6
26 27 1 58
26 27 NA 180
26 28 2 30
26 29 3 9
26 30 4 3
26 31 5 1
27 27 0 3
27 28 1 45
27 28 NA 131
27 29 2 29
27 30 3 11
27 31 4 4
27 32 5 3
27 33 6 1
27 34 7 1
28 28 0 5
28 29 1 46
28 29 NA 100
28 30 2 17
28 31 3 10
28 32 4 3
28 33 5 1
29 29 0 3
29 30 1 32
29 30 NA 89
29 31 2 11
29 32 3 2
29 33 4 5
29 35 6 1
30 30 0 1
30 31 1 18
30 31 NA 60
30 32 2 11
30 33 3 1
31 31 0 4
31 32 1 15
31 32 NA 42
31 33 2 6
31 34 3 5
31 35 4 2
32 32 0 1
32 33 1 12
32 33 NA 49
32 34 2 8
32 35 3 1
33 33 0 1
33 34 1 14
33 34 NA 41
33 35 2 7
34 35 1 11
34 35 NA 26
34 36 2 1
34 37 3 2
34 39 5 1
35 35 0 1
35 36 1 10
35 36 NA 19
35 37 2 4
35 38 3 1
36 36 0 1
36 37 1 5
36 37 NA 11
36 38 2 3
37 37 0 1
37 38 1 3
37 38 NA 11
37 39 2 2
37 40 3 1
37 41 4 1
38 39 1 2
38 39 NA 7
38 40 2 4
38 42 4 1
39 40 1 2
39 40 NA 5
39 43 4 1
40 41 1 1
40 41 NA 2
40 42 2 1
41 42 NA 3
41 44 3 1
42 43 1 4
42 43 NA 5
43 44 NA 1
43 45 2 1
44 45 NA 1
44 47 3 1
45 46 1 1
45 46 NA 3
46 47 NA 2
47 48 NA 2
49 50 NA 1
51 52 NA 1
56 57 NA 1
86 87 NA 1
88 89 NA 2
NA NA NA 7359
Table 44: PD196: Animal naming
Characteristic N = 20,9121
Animal naming (correct - errors)
    0 68 (0.5%)
    1 35 (0.3%)
    2 70 (0.5%)
    3 109 (0.8%)
    4 162 (1.2%)
    5 257 (1.9%)
    6 335 (2.5%)
    7 396 (2.9%)
    8 504 (3.7%)
    9 511 (3.8%)
    10 621 (4.6%)
    11 657 (4.8%)
    12 697 (5.1%)
    13 693 (5.1%)
    14 763 (5.6%)
    15 745 (5.5%)
    16 756 (5.6%)
    17 683 (5.0%)
    18 690 (5.1%)
    19 667 (4.9%)
    20 635 (4.7%)
    21 534 (3.9%)
    22 488 (3.6%)
    23 451 (3.3%)
    24 411 (3.0%)
    25 304 (2.2%)
    26 287 (2.1%)
    27 228 (1.7%)
    28 182 (1.3%)
    29 143 (1.1%)
    30 91 (0.7%)
    31 74 (0.5%)
    32 71 (0.5%)
    33 63 (0.5%)
    34 41 (0.3%)
    35 35 (0.3%)
    36 20 (0.1%)
    37 19 (0.1%)
    38 14 (0.1%)
    39 8 (<0.1%)
    40 4 (<0.1%)
    41 4 (<0.1%)
    42 9 (<0.1%)
    43 2 (<0.1%)
    44 2 (<0.1%)
    45 4 (<0.1%)
    46 2 (<0.1%)
    47 2 (<0.1%)
    49 1 (<0.1%)
    51 1 (<0.1%)
    56 1 (<0.1%)
    86 1 (<0.1%)
    88 2 (<0.1%)
    Unknown 7,359
1 n (%)

Word recall

I found two variables PD174 and PD184 that are the number of words recalled correctly, so we don’t need to recreate these scores with PD182M1-PD182M13 and PD183M1-PD183M13.

No recoding was done.

Are we going to use both immediate and delayed recall? Or just delayed recall?

Table 45: vdwdimm: Immediate word recall recoding
Word recall - Immediate NUMBER GOOD - IMMEDIATE n
0 0 181
1 1 172
2 2 600
3 3 1574
4 4 3321
5 5 4712
6 6 4534
7 7 2912
8 8 1324
9 9 426
10 10 109
NA NA 1047
Table 46: vdwdimm: Immediate word recall
Characteristic N = 20,9121
Word recall - Immediate
    0 181 (0.9%)
    1 172 (0.9%)
    2 600 (3.0%)
    3 1,574 (7.9%)
    4 3,321 (17%)
    5 4,712 (24%)
    6 4,534 (23%)
    7 2,912 (15%)
    8 1,324 (6.7%)
    9 426 (2.1%)
    10 109 (0.5%)
    Unknown 1,047
1 n (%)
Table 47: vdwddel: Delayed word recall recoding
Word recall - Delayed NUMBER GOOD - DELAYED n
0 0 1255
1 1 815
2 2 1517
3 3 2847
4 4 3986
5 5 4107
6 6 2942
7 7 1472
8 8 627
9 9 234
10 10 62
NA NA 1048
Table 48: vdwddel: Delayed word recall
Characteristic N = 20,9121
Word recall - Delayed
    0 1,255 (6.3%)
    1 815 (4.1%)
    2 1,517 (7.6%)
    3 2,847 (14%)
    4 3,986 (20%)
    5 4,107 (21%)
    6 2,942 (15%)
    7 1,472 (7.4%)
    8 627 (3.2%)
    9 234 (1.2%)
    10 62 (0.3%)
    Unknown 1,048
1 n (%)

Numeracy

Recoded 996, 997, and 999 as missing.

Table 49: vdexf7: Number series score recoding
Number series CALCULATED NUMBER SERIES SCORE n
409 409 159
413 413 124
429 429 226
435 435 112
462 462 424
465 465 279
484 484 251
485 485 216
488 488 97
489 489 862
501 501 1132
503 503 1069
513 513 1058
514 514 830
518 518 1028
519 519 809
524 524 376
525 525 375
528 528 720
529 529 897
536 536 1254
537 537 1430
546 546 984
547 547 229
549 549 714
558 558 937
567 567 519
570 570 285
584 584 449
NA 996 945
NA 997 915
NA 999 258
NA NA 949
Table 50: vdexf7: Number series score
Characteristic N = 20,9121
Number series
    409 159 (0.9%)
    413 124 (0.7%)
    429 226 (1.3%)
    435 112 (0.6%)
    462 424 (2.4%)
    465 279 (1.6%)
    484 251 (1.4%)
    485 216 (1.2%)
    488 97 (0.5%)
    489 862 (4.8%)
    501 1,132 (6.3%)
    503 1,069 (6.0%)
    513 1,058 (5.9%)
    514 830 (4.7%)
    518 1,028 (5.8%)
    519 809 (4.5%)
    524 376 (2.1%)
    525 375 (2.1%)
    528 720 (4.0%)
    529 897 (5.0%)
    536 1,254 (7.0%)
    537 1,430 (8.0%)
    546 984 (5.5%)
    547 229 (1.3%)
    549 714 (4.0%)
    558 937 (5.3%)
    567 519 (2.9%)
    570 285 (1.6%)
    584 449 (2.5%)
    Unknown 3,067
1 n (%)

Rescale cognitive items

The continuous cognitive items were rescaled using the min/max normalization.

Table 51: Descriptive statistics of cognitive items after rescaling
Characteristic N = 20,912
Animal naming (correct - errors)
    N Non-missing 13,553
    Mean (SD) 0.18 (0.08)
    Median (Q1, Q3) 0.18 (0.13, 0.24)
    Min, Max 0.00, 1.00
    Unknown 7,359
Word recall - Immediate
    N Non-missing 19,865
    Mean (SD) 0.53 (0.17)
    Median (Q1, Q3) 0.50 (0.40, 0.60)
    Min, Max 0.00, 1.00
    Unknown 1,047
Word recall - Delayed
    N Non-missing 19,864
    Mean (SD) 0.43 (0.20)
    Median (Q1, Q3) 0.40 (0.30, 0.60)
    Min, Max 0.00, 1.00
    Unknown 1,048
Number series
    N Non-missing 17,845
    Mean (SD) 0.64 (0.18)
    Median (Q1, Q3) 0.66 (0.54, 0.73)
    Min, Max 0.00, 1.00
    Unknown 3,067

Model results

A single factor CFA model was fit to the items.

Model 1

The first model used all the items.

The model fit for this model was “poor”.

Table 52: Model 1: Fit statistics
Fit statistic Value
RMSEA : Estimate 0.087
CFI 0.903
SRMR 0.057
Table 53: Model 1: Factor loadings
Item Label Std Factor Loading
vdori Orientation to time 0.484
vdlfl1z Animal naming 0.593
vdlfl2 Scissors & cactus 0.661
vdlfl3 President & vice-president 0.581
vdwdimmz Immediate word recall 0.686
vdwddelz Delayed word recall 0.681
vdexf7z Number series 0.613
vdsevens Serial sevens 0.664
vdcount Count backwards from 20 0.545

Model 2

The second model removed the immediate word recall item.

The model fit for this model was “good”.

Table 54: Model 2: Fit statistics
Fit statistic Value
RMSEA : Estimate 0.057
CFI 0.961
SRMR 0.041
Table 55: Model 2: Factor loadings
Item Label Std Factor Loading
vdori Orientation to time 0.485
vdlfl1z Animal naming 0.591
vdlfl2 Scissors & cactus 0.678
vdlfl3 President & vice-president 0.599
vdwddelz Delayed word recall 0.589
vdexf7z Number series 0.642
vdsevens Serial sevens 0.706
vdcount Count backwards from 20 0.567

Factor scores

This is the distribution of the factor scores from the final model.

Figure 3: Distribution of factor scores by HCAP inclusion status
Figure 4: Distribution of factor scores by HCAP inclusion status, including HRS sample

Norming the factor score

The HRS sample was filtered to those that were used as the norming sample in HCAP.

The demographic variables were centered to the mean values in the norming sample. The same centering values were used in the full HRS sample.

These are the centering values used:

Table 56: Demographic variable means
variable mean
female 0.601
black 0.148
hisp 0.100
SCHLYRS 13.329

Summary of the centered demographic variables in the norming sample:

Table 57: Descriptive statistics of centered demographic variables - HCAP
Characteristic N = 1,7871
Female (centered to HCAP normal sample)
    -0.601007274762173 713 / 1,787 (40%)
    0.398992725237827 1,074 / 1,787 (60%)
Black (centered to HCAP normal sample)
    -0.148293228875209 1,522 / 1,787 (85%)
    0.851706771124791 265 / 1,787 (15%)
Hispanic (centered to HCAP normal sample)
    -0.0996082820369341 1,609 / 1,787 (90%)
    0.900391717963066 178 / 1,787 (10.0%)
School years (centered to HCAP normal sample) 0.00 (2.81)
    Unknown 2
1 n / N (%); Mean (SD)

Summary of the centered demographic variables in the full HRS sample:

Table 58: Descriptive statistics of centered demographic variables - HRS
Characteristic N = 20,9121
Female (centered to HCAP normal sample)
    -0.601007274762173 8,664 / 20,909 (41%)
    0.398992725237827 12,245 / 20,909 (59%)
    Unknown 3
Black (centered to HCAP normal sample)
    -0.148293228875209 16,475 / 20,909 (79%)
    0.851706771124791 4,434 / 20,909 (21%)
    Unknown 3
Hispanic (centered to HCAP normal sample)
    -0.0996082820369341 17,482 / 20,909 (84%)
    0.900391717963066 3,427 / 20,909 (16%)
    Unknown 3
School years (centered to HCAP normal sample) -0.6 (3.3)
    Unknown 91
1 n / N (%); Mean (SD)

Age was modeled with a cubic regression spline. The placement of the knots was determined in the norming sample, using the default percentiles of 5%, 35%, 65%, and 95%. These same knots were used in creating the splines in the full HRS sample.

Theses are the knots used:

 5% 35% 65% 95% 
 65  70  76  85 

To double check the math, these splines were compared to those from the rms::rcs() function.

These are the first 10 rows from the manually created splines.

Table 59: Manually created regression splines
Age spline 1 (from HCAP normal sample) Age spline 2 (from HCAP normal sample) Age spline 3 (from HCAP normal sample)
70 0.312 0.000
79 6.710 1.710
79 6.710 1.710
85 15.950 5.400
77 4.314 0.853
77 4.314 0.853
75 2.500 0.312
77 4.314 0.853
69 0.160 0.000
88 20.900 7.425

These are the first 10 rows from the rms::rcs() function.

Table 60: rcs() created regression splines
hrs16_cog_norm hrs16_cog_norm' hrs16_cog_norm''
70 0.313 0.000
79 6.710 1.710
79 6.710 1.710
85 15.950 5.400
77 4.314 0.853
77 4.314 0.853
75 2.500 0.313
77 4.314 0.853
69 0.160 0.000
88 20.900 7.425

Factor Score Transformation

The factor score was transformed using a Blom transformation. The transformation was done in the norming sample. Cubic regression splines of the factor score were created to predict the Blom score from the factor score. These cubic regression splines were recreated in the HRS sample to get predicted Blom scores in the HRS sample.

The knots used to create the cubic splines for the Factor score are:

      5%      35%      65%      95% 
-1.18125 -0.09350  0.50775  1.29150 

The regression model is able to predict the blom score well using the factor score cubic splines.

Figure 5: Original Blom score vs Predicted Blom score in norming sample

The cubic splines using the same knots were recreated in the full HRS sample and were used to predict the blom transformed factor score.

Normalization of factor scores

The factor scores were normalized by regressing the predicted blom score on the demographic variables described above in the norming sample. Two way interactions between the variables were included.

Table 61: Parameter estimates of the normalization regression in the norming sample
term estimate std.error statistic p.value
(Intercept) 0.029 1.182 0.024 0.981
spage1 0.002 0.018 0.123 0.902
spage2 −0.138 0.077 −1.792 0.073
spage3 0.297 0.174 1.709 0.088
cfemale −3.593 2.358 −1.524 0.128
cblack −1.828 3.186 −0.574 0.566
chisp −4.370 3.896 −1.122 0.262
cschlyrs 0.233 0.445 0.523 0.601
spage1:cfemale 0.053 0.035 1.495 0.135
spage1:cblack 0.015 0.048 0.307 0.759
spage1:chisp 0.055 0.058 0.936 0.349
spage1:cschlyrs −0.001 0.007 −0.143 0.886
spage2:cfemale −0.194 0.155 −1.257 0.209
spage2:cblack −0.024 0.214 −0.113 0.910
spage2:chisp −0.176 0.280 −0.628 0.530
spage2:cschlyrs 0.010 0.030 0.344 0.731
spage3:cfemale 0.421 0.348 1.210 0.226
spage3:cblack 0.023 0.488 0.048 0.962
spage3:chisp 0.344 0.650 0.529 0.597
spage3:cschlyrs −0.033 0.067 −0.499 0.618
cfemale:cblack 0.160 0.116 1.386 0.166
cfemale:chisp −0.082 0.139 −0.585 0.559
cfemale:cschlyrs 0.023 0.014 1.622 0.105
cblack:cschlyrs 0.010 0.021 0.472 0.637
chisp:cschlyrs −0.084 0.018 −4.708 0.000

Plots of observed vs model implied blom scores

These are plots of the observed blom scores versus the model implied blom scores that adjusted for demographic characteristics.

Figure 6: Predicted blom vs normalized blom in norming sample

The regression model was used to estimate the normalized blom scores in the full HRS sample.

Figure 7: Predicted blom vs normalized blom in full HRS sample

Transforming to T-scores

The normalized blom scores were transformed to a T-score using the following equation:

T= 50 + 10 * ((Predicted blom score - Normalized blom score)/(sd(predicted blom score)*sqrt(1-r2)))

The sd of the predicted blom score in the norming sample is: 0.999.

The adjusted model r2 is 0.389 .

Table 62: Summary statistics of the factor score variables in the norming sample
Characteristic N = 1,7871
Predicted Blom score (model from HCAP normal sample) 0.00 (1.00)
    Unknown 11
Adjusted Blom score (from to HCAP normal sample) 0.00 (0.63)
    Unknown 2
T-scaled F score (from to HCAP normal sample) 50 (10)
    Unknown 13
1 Mean (SD)
Figure 8: Predicted blom score vs T-score in norming sample
Table 63: Summary statistics of the factor score variables in the full HRS sample
Characteristic N = 20,9121
Predicted Blom score (model from HCAP normal sample) -0.20 (1.10)
    Unknown 942
Adjusted Blom score (from to HCAP normal sample) -0.16 (0.82)
    Unknown 91
T-scaled F score (from to HCAP normal sample) 49 (12)
    Unknown 1,030
1 Mean (SD)
Figure 9: Predicted blom score vs T-score in the full HRS sample

Designing the algorithm

This section details how we design the algorithm to mimic the Manly-Jones (2022) algorithm that we implemented in HCAP. In the previous sections we created the components that go into the algorithm - the cognitive factor score, the functional impairment measure, the subjective cognitive complaints indicator. However, these components are different than the ones used in HCAP. So, we need to find cutpoints for these components that give similar prevalence as HCAP.

These are the diagrams of the HCAP algorithm and the proposed HRS algorithm. In both algorithms the fist step is to decide if a participant has normal cognition, mild impairment, or severe impairment. After finding the cutoffs to use to determine cognitive impairment level, the next step is to find cutoffs to determine functional impairment. The functional impairment cutoffs will be determined separately for each level of cognitive impairment.

HCAP algorithm
Figure 10: HCAP algorithm
HRS algorithm
Figure 11: HRS algorithm

Finding cognitive impairment thresholds

Figure 12 shows the cognitive factor scores are highly correlated between the HCAP cognitive battery and the HRS cognitive battery. The correlation between them is 0.697.

Figure 12: Cognitive factor score in HCAP vs HRS

Figure 13 shows the distribution of the HCAP cognitive factor score by the number of cognitive domains impaired. The plot shows there is good separation in the distribution of the HCAP factor scores by the number of domains impaired.

Figure 13: Distribution of HCAP cognitive factor score by number of domains impaired

Figure 14 is a similar figure that shows the distribution of the HRS cognitive factor score by the number of HCAP cognitive domains impaired. This figure shows there is more overlap of the distributions of the factor scores when separating them by the same variable as in Figure 13.

Figure 14: Distribution of HRS cognitive factor score by number of domains impaired

Table 64 shows that 19.6% had one domain impaired, and 15.7% had two or more domains impaired. So we’ll find cutpoints in the HRS factor score to match these percentages.

Table 64: Distribution of the number of impaired domains in HCAP
Characteristic N = 2,9931
Number of domains impaired
    No domains 1,935 (65%)
    1 domain 587 (20%)
    2+ domains 471 (16%)
1 n (%)

The values of the HRS factor score that match those percentages are:

     15%      35% 
36.43890 43.77599 
Table 65: Level of cognitive impairment in HRS matched to the number of domains impaired
Characteristic N = 2,9931
Level of cognitive impairment (HRS)
    None 1,941 (67%)
    Mild 572 (20%)
    Severe 405 (14%)
    Unknown 75
1 n (%)

Table 66 shows that the agreement of level of cognitive impairment is high between the two measures. But despite choosing thresholds of the factor score to match the percentages of the number of impaired domains, the agreement is high but not perfect. The weighted kappa is 0.49.

Table 66: Crosstab of number of domains impaired (HCAP) vs matched factor scores (HRS)
Level of cognitive impairment (HRS)
Total
None Mild Severe Unknown
Number of domains impaired




    No domains 1,528 307 88 12 1,935
    1 domain 312 164 99 12 587
    2+ domains 101 101 218 51 471
Total 1,941 572 405 75 2,993

Finding functional impairment thresholds

The next step of the algorithm after finding cognitive impairment thresholds is to find thresholds to for functional impairment. This will be done separately by cognitive impairment level.

In the HCAP algorithm, in the 2 or more impaired domains category, 391 / (391 + 211) = 65% had functional impairments. If we use a threshold of 1 or more ADL/IADL impairments then 53.6 % will be categorized as having funcitonal impairment.

Table 67: Distribution of functional impairment among those with severve cognitive impairment
Characteristic N = 4051
Sum of ADL/IADL impairments
    0 188 (46%)
    1 97 (24%)
    2 36 (8.9%)
    3 29 (7.2%)
    4 20 (4.9%)
    5 8 (2.0%)
    6 17 (4.2%)
    7 4 (1.0%)
    8 1 (0.2%)
    9 3 (0.7%)
    10 2 (0.5%)
1 n (%)

In the HCAP algorithm, in the 1 impaired domain category, 536 / (536 + 231) = 70% had functional impairments. If we use a threshold of 1 or more ADL/IADL impairments then 35.8 % will be categorized as having funcitonal impairment.

Table 68: Distribution of functional impairment among those with mild cognitive impairment
Characteristic N = 5721
Sum of ADL/IADL impairments
    0 367 (64%)
    1 98 (17%)
    2 54 (9.4%)
    3 20 (3.5%)
    4 16 (2.8%)
    5 8 (1.4%)
    6 6 (1.0%)
    7 3 (0.5%)
1 n (%)

Informant based cognitive impairment

Table 69: Crosstab of missing cognitive score and missing informant score
TF_missing
Total
0 1
jorm_missing


    0 0 71 71
    1 2,918 4 2,922
Total 2,918 75 2,993

Apply algorithm in the HCAP sample

In the previous section we found the cutpoints to use to determine cognitive impairment and functional impairment levels. In this section we use those cutpoints to implement the algorithm.

The cutpoints we used for cognitive impairment were less than 36.0 and less than 43.3. The ADL/IADL cutpoints were 1 or more impairments for both of the cognitive levels. The Jorm thresholds were less than 3.0 and less than 3.4.

The following code shows a function that uses these thresholds to create the data file.

algorithm_thresholds
function(df){
  df <- df %>%
    mutate(cog_threshold = case_when(TF <36.0 ~ 2,
                                     TF < 43.3 ~ 1,
                                     !is.na(TF) ~ 0),
           iadl_threshold = case_when(cog_threshold==2 & iadl_imp>0 ~ 1,
                                      cog_threshold==2 & iadl_imp==0 ~ 0,
                                      cog_threshold==1 & iadl_imp>0 ~ 1,
                                      cog_threshold==1 & iadl_imp==0 ~ 0),
           jorm_threshold = case_when(vs3jormsc>= 3.4 ~ 2,
                                      vs3jormsc > 3.0 &  vs3jormsc < 3.4 ~ 1,
                                      vs3jormsc <= 3.0  ~ 0)
    )

  df
}
<bytecode: 0x164cda6d8>

Version 1 of algorithm

Figure 15 shows the first version of the algorithm.

Figure 15: HRS algorithm (v1)

This is the code that creates a function that implements the algortihm shown in Figure 15.

Note: The algorithm uses the Jorm score from HCAP, not the HRS version.

v1_algorithm
function(df) {
  df <- df %>%
    mutate(dx_v1 = case_when(
      cog_threshold == 2 & iadl_threshold == 1 ~ 2,
      cog_threshold == 2 & iadl_threshold == 0 & as.numeric(haven::zap_labels(self_concerns)) == 1 ~ 2,
      cog_threshold == 2 & iadl_threshold == 0 & as.numeric(haven::zap_labels(self_concerns)) == 0 ~ 1,
      cog_threshold == 1 & iadl_threshold == 1 ~ 1,
      cog_threshold == 1 & iadl_threshold == 0 & as.numeric(haven::zap_labels(self_concerns)) == 1 ~ 1,
      cog_threshold == 1 & iadl_threshold == 0 & as.numeric(haven::zap_labels(self_concerns)) == 0 ~ 0,
      cog_threshold == 0 ~ 0,
      is.na(TF) & jorm_threshold == 2 ~ 2,
      is.na(TF) & jorm_threshold == 1 ~ 1,
      is.na(TF) & jorm_threshold == 0 ~ 0
    ))

  df <- df %>%
    labelled::set_variable_labels(dx_v1 = "Algorithm (V1) Dx") %>%
    labelled::set_value_labels(dx_v1 = c("Normal" = 0, "MCI" = 1, "Dementia" = 2))

  df
}
<bytecode: 0x164d06e10>
Table 70: Relationship between version 1 diagnosis and its constituent components
Algorithm (V1) Dx cog_threshold iadl_threshold Compared to two years ago, would you say your memory is better now, about the same, or worse now than it was then? jorm_threshold n
0 0 NA 0 0 927
0 0 NA 0 1 428
0 0 NA 0 2 108
0 0 NA 1 0 228
0 0 NA 1 1 171
0 0 NA 1 2 79
0 1 0 0 0 157
0 1 0 0 1 84
0 1 0 0 2 29
0 NA NA 0 0 1
0 NA NA NA 0 8
1 1 0 1 0 34
1 1 0 1 1 40
1 1 0 1 2 22
1 1 1 0 0 50
1 1 1 0 1 36
1 1 1 0 2 28
1 1 1 1 0 27
1 1 1 1 1 26
1 1 1 1 2 38
1 2 0 0 0 66
1 2 0 0 1 32
1 2 0 0 2 29
1 NA NA 0 1 3
1 NA NA NA 1 12
2 2 0 1 0 18
2 2 0 1 1 18
2 2 0 1 2 25
2 2 1 0 0 49
2 2 1 0 1 26
2 2 1 0 2 62
2 2 1 1 0 8
2 2 1 1 1 23
2 2 1 1 2 48
2 2 1 NA 2 1
2 NA NA NA 2 51
NA 1 0 NA 0 1

The following table shows the sample sizes at the places labeled in Figure 15.

Table 71: Sample sizes at various places in the flow diagram
address n
a1 75
a2 2513
b2 405
b1 24
b3 9
c4 15
c1 51
b4 572
c2 217
c3 188
d1 61
d2 127
c5 205
c6 367
d3 96
d4 270
c7 1941
e1 0
e2 329
e3 443
e4 2220
Table 72: Distribution of dementia diagnoses in HCAP
Characteristic N = 2,9931
Algorithm (V1) Dx
    Normal 2,220 (74%)
    MCI 443 (15%)
    Dementia 329 (11%)
    Unknown 1
HRS HCAP Dementia and MCI Classification (EAP version, HRS HCAP variable set 1)
    Normal (1) 2,059 (69%)
    MCI (2) 652 (22%)
    Dementia (3) 282 (9.4%)
1 n (%)
Table 73: Crosstab of version 1 diagnosis and HCAP diagnosis
Algorithm (V1) Dx
Total
Normal MCI Dementia Unknown
HRS HCAP Dementia and MCI Classification (EAP version, HRS HCAP variable set 1)




    Normal (1) 1,789 201 68 1 2,059
    MCI (2) 374 168 110 0 652
    Dementia (3) 57 74 151 0 282
Total 2,220 443 329 1 2,993
$kappa
[1] 0.3390473

$weighted.kappa
[1] 0.5238401

Version 2

In version 2 of the algorithm we removed the self-rated concerns from the severe cognitive impairment portion of the algorithm. Now, if a participant has severe cognitive impairment but no functional impairment, they will be classified as MCI.

Figure 16: HRS algorithm (v2)

This is the code that creates a function that implements the algortihm shown in Figure 16.

v2_algorithm
function(df) {
  df <- df %>%
    mutate(dx_v2 = case_when(
      cog_threshold == 2 & iadl_threshold == 1 ~ 2,
      cog_threshold == 2 & iadl_threshold == 0 ~ 1,
      cog_threshold == 1 & iadl_threshold == 1 ~ 1,
      cog_threshold == 1 & iadl_threshold == 0 & as.numeric(haven::zap_labels(self_concerns)) == 1 ~ 1,
      cog_threshold == 1 & iadl_threshold == 0 & as.numeric(haven::zap_labels(self_concerns)) == 0 ~ 0,
      cog_threshold == 0 ~ 0,
      is.na(TF) & jorm_threshold == 2 ~ 2,
      is.na(TF) & jorm_threshold == 1 ~ 1,
      is.na(TF) & jorm_threshold == 0 ~ 0
    ))

  df <- df %>%
    labelled::set_variable_labels(dx_v2 = "Algorithm (V2) Dx") %>%
    labelled::set_value_labels(dx_v2 = c("Normal" = 0, "MCI" = 1, "Dementia" = 2))

  df
}
<bytecode: 0x164d642b0>
Table 74: Relationship between version 2 diagnosis and its constituent components
Algorithm (V2) Dx cog_threshold iadl_threshold Compared to two years ago, would you say your memory is better now, about the same, or worse now than it was then? jorm_threshold n
0 0 NA 0 0 927
0 0 NA 0 1 428
0 0 NA 0 2 108
0 0 NA 1 0 228
0 0 NA 1 1 171
0 0 NA 1 2 79
0 1 0 0 0 157
0 1 0 0 1 84
0 1 0 0 2 29
0 NA NA 0 0 1
0 NA NA NA 0 8
1 1 0 1 0 34
1 1 0 1 1 40
1 1 0 1 2 22
1 1 1 0 0 50
1 1 1 0 1 36
1 1 1 0 2 28
1 1 1 1 0 27
1 1 1 1 1 26
1 1 1 1 2 38
1 2 0 0 0 66
1 2 0 0 1 32
1 2 0 0 2 29
1 2 0 1 0 18
1 2 0 1 1 18
1 2 0 1 2 25
1 NA NA 0 1 3
1 NA NA NA 1 12
2 2 1 0 0 49
2 2 1 0 1 26
2 2 1 0 2 62
2 2 1 1 0 8
2 2 1 1 1 23
2 2 1 1 2 48
2 2 1 NA 2 1
2 NA NA NA 2 51
NA 1 0 NA 0 1

The following table shows the sample sizes at the places labeled in Figure 16.

Table 75: Sample sizes at various places in the v2 flow diagram
address n
a1 75
a2 2513
b1 405
b2 572
b3 1941
c1 24
c2 9
d1 51
d2 217
d3 188
d4 15
d5 205
d6 367
e1 96
e2 270
f1 0
f2 268
f3 504
f4 2220
Table 76: Distribution of the diagnoses for the subset that was affected by the change in algorithm
Characteristic N = 611
Algorithm (V1) Dx
    Normal 0 (0%)
    MCI 0 (0%)
    Dementia 61 (100%)
Algorithm (V2) Dx
    Normal 0 (0%)
    MCI 61 (100%)
    Dementia 0 (0%)
HRS HCAP Dementia and MCI Classification (EAP version, HRS HCAP variable set 1)
    Normal (1) 15 (25%)
    MCI (2) 22 (36%)
    Dementia (3) 24 (39%)
1 n (%)
Table 77: Distribution of dementia diagnoses in HCAP
Characteristic N = 2,9931
Algorithm (V2) Dx
    Normal 2,220 (74%)
    MCI 504 (17%)
    Dementia 268 (9.0%)
    Unknown 1
HRS HCAP Dementia and MCI Classification (EAP version, HRS HCAP variable set 1)
    Normal (1) 2,059 (69%)
    MCI (2) 652 (22%)
    Dementia (3) 282 (9.4%)
1 n (%)
Table 78: Crosstab of version 1 diagnosis and HCAP diagnosis
Algorithm (V1) Dx
Total
Normal MCI Dementia Unknown
HRS HCAP Dementia and MCI Classification (EAP version, HRS HCAP variable set 1)




    Normal (1) 1,789 201 68 1 2,059
    MCI (2) 374 168 110 0 652
    Dementia (3) 57 74 151 0 282
Total 2,220 443 329 1 2,993
Table 79: Crosstab of version 2 diagnosis and HCAP diagnosis
Algorithm (V2) Dx
Total
Normal MCI Dementia Unknown
HRS HCAP Dementia and MCI Classification (EAP version, HRS HCAP variable set 1)




    Normal (1) 1,789 216 53 1 2,059
    MCI (2) 374 190 88 0 652
    Dementia (3) 57 98 127 0 282
Total 2,220 504 268 1 2,993
$kappa
[1] 0.3337944

$weighted.kappa
[1] 0.5156637
Table 80: Crosstab of version 2 diagnosis and version 1 diagnosis
Algorithm (V2) Dx
Total
Normal MCI Dementia Unknown
Algorithm (V1) Dx




    Normal 2,220 0 0 0 2,220
    MCI 0 443 0 0 443
    Dementia 0 61 268 0 329
    Unknown 0 0 0 1 1
Total 2,220 504 268 1 2,993